ANRV374-EA37-10
ARI
ANNUAL
REVIEWS
27 March 2009
14:57
Further
Annu. Rev. Earth Planet. Sci. 2009.37:209-236. Downloaded from arjournals.annualreviews.org
by University of British Columbia Library on 12/29/09. For personal use only.
Click here for quick links to
Annual Reviews content online,
including:
rOther articles in this volume
r Top cited articles
r Top downloaded articles
r0VSDPNQrehensive search
Forensic Seismology
and the Comprehensive
Nuclear-Test-Ban Treaty
David Bowers and Neil D. Selby
AWE Blacknest, Brimpton, Reading RG7 4RS, United Kingdom;
email: [email protected], [email protected]
Annu. Rev. Earth Planet. Sci. 2009. 37:209–36
Key Words
First published online as a Review in Advance on
January 20, 2009
global seismology, International Monitoring System, seismic
discrimination methods, underground explosion source, earthquake source
The Annual Review of Earth and Planetary Sciences is
online at earth.annualreviews.org
This article’s doi:
10.1146/annurev.earth.36.031207.124143
c 2009 by Annual Reviews.
Copyright !
All rights reserved
0084-6597/09/0530-0209$20.00
Abstract
One application of forensic seismology is to help verify compliance with
the Comprehensive Nuclear-Test-Ban Treaty. One of the challenges facing
the forensic seismologist is to discriminate between the many thousands of
earthquakes of potential interest each year and potential Treaty violations
(underground explosions). There are four main methods: (a) ratio of bodyto surface-wave magnitudes, (b) ratio of high-frequency P to S energy,
(c) model-based methods, and (d ) source depth. Methods (a) and (b) have an
empirical basis. The weakness of methods (a)–(c) is the lack of an equivalent
elastic source for an underground explosion fired in the range of geological
media found around the world. Reliable routine source-depth determination
has proved difficult. However, experience gained in the past decade at identifying suspicious seismic sources suggests that although no single method
works all of the time, intelligent and original application of complementary
methods is usually sufficient to satisfactorily identify the source in question.
209
ANRV374-EA37-10
ARI
27 March 2009
14:57
INTRODUCTION
Global seismology is the study of elastic (seismic) waves to investigate a seismic source and the
structure and processes within Earth. Usually, the seismic waves studied are caused by earthquakes,
but seismic waves are also generated by nuclear explosions. It was recognized by experts, meeting
in Geneva in 1958, that seismology could be the only way to detect and identify underground
explosions, and thus could be used to help verify a treaty banning nuclear-test explosions.
CTBT:
Comprehensive
Nuclear-Test-Ban
Treaty
IMS: International
Monitoring System
Seismology and the Comprehensive Nuclear-Test-Ban Treaty
Annu. Rev. Earth Planet. Sci. 2009.37:209-236. Downloaded from arjournals.annualreviews.org
by University of British Columbia Library on 12/29/09. For personal use only.
OSI: On-Site
Inspection
The Comprehensive Nuclear-Test-Ban Treaty (CTBT) took more than 40 years to achieve and is
considered to be the Holy Grail by arms control advocates (Marshall et al. 2001). The verification
system for the CTBT has four pillars: the International Monitoring System (IMS), Consultation
and Clarification, On-Site Inspection (OSI), and Confidence Building Measures. The IMS will
consist of a global network of more than 300 stations sending data to the International Data Center
(IDC) in Vienna, Austria.
The technical discussions on whether a CTBT could be adequately verified, which spanned
the period from 1958 to the final negotiation of the Treaty in 1996, have been extensively reviewed elsewhere (Blandford 1977; Douglas 1981, 2007)—the main technical difficulty was the
development of adequate seismological methods of identifying underground explosions (potential
Treaty violations) from the many thousands of earthquakes that occur each year. The relative importance of seismology for Treaty verification is reflected by the fact that more than half (170 out
of 321) of the IMS stations will be seismic stations. However, seismology alone cannot distinguish
between single nuclear and chemical underground explosions—for this, diagnostic radionuclides
are required. To detect and attribute the diagnostic radionuclides, the IMS network will comprise
80 radionuclide stations, and the Treaty has provisions for OSI.
Here, we use the term forensic seismology to describe seismology applied to CTBT verification
using data mainly from global networks of seismic stations. The search for aftershocks from a
suspected underground nuclear explosion as part of an OSI (Takano & Krioutchenkov 2001) could
also be considered forensic seismology—examples of other applications of forensic seismology are
briefly described in the sidebar Other Applications of Forensic Seismology. [The use of the term
forensic comes from its definition as “pertaining to courts of justice.” Forensic seismology is thus
the application of seismological science to the elucidation of doubtful questions in such a court
(Douglas 2007).]
IDC: International
Data Center
The IMS and Seismology
The IMS network of stations will comprise 50 primary seismic stations providing continuous data
and will be supported by 120 auxiliary seismic stations that provide data on request (Barrientos
et al. 2001), 11 hydroacoustic stations (Lawrence et al. 2001), 60 infrasound stations (Christie et al.
2001), and 80 radionuclide stations (with particulate and noble gas sensors) (Matthews & Schulze
2001). The radionuclide part of the IMS will also be supported by 16 radionuclide laboratories
(Karhu & Clawson 2001). The IDC receives data from the IMS network of stations and distributes
these data and derived products (e.g., bulletins of seismic events) to states that are party to the
Treaty (Bratt 2001).
At the IDC, the continuous waveform data (primary seismic, hydroacoustic, and infrasound)
are searched for signals from a potential Treaty violation, and lists of detections are formed. These
detections are then associated with events, and the event location, depth, and origin time (and
other associated parameters, such as magnitude) are estimated (Bratt 2001). In general, the seismic
signals are searched for signals from an underground explosion, the hydroacoustic for an explosion
210
Bowers
·
Selby
ANRV374-EA37-10
ARI
27 March 2009
14:57
OTHER APPLICATIONS OF FORENSIC SEISMOLOGY
The rapid expansion of seismometer observatories around the world has resulted in a growing number of examples
of how seismic observations have been used to provide evidence to support investigations that are potentially judicial.
Annu. Rev. Earth Planet. Sci. 2009.37:209-236. Downloaded from arjournals.annualreviews.org
by University of British Columbia Library on 12/29/09. For personal use only.
1. Seismic and acoustic signals (both recorded by a seismometer) from aircraft crashes can provide independent
estimates of impact time and location, for example, the impacts from Pan Am Flight 103 at Lockerbie, Scotland
in 1988 (crash site approximately 25 km southwest of the Eskdalemuir seismometer array), the impacts from Swiss
Air passenger aircraft in shallow water at Peggy’s Cove near Halifax, Canada, in 1998 (McCormack 2003), and the
aircraft impacts into (and subsequent collapse of ) the twin towers of the World Trade Center on September 11,
2001 (Kim et al. 2001).
2. Seismic and acoustic signals can provide independent estimates of the detonation time of terrorist explosions,
such as from the Oklahoma City bombing in 1993 (Holzer et al. 1996) and from the truck bomb attack on the US
Embassy in Nairobi, Kenya, in 1998 (Koper et al. 1999).
3. The amplitude of seismic and acoustic signals can be used to place constraints on the explosive yield (and coupling)
of potentially judicial explosive sources (Koper et al. 1999). Experiments using controlled truck-bomb sources have
refined initial estimates of the yield of the Nairobi bomb (equivalent to 2 to 6 tons of TNT) and resulted in initial
scaling laws that could be used in future investigations (Koper et al. 2002).
in the ocean and the infrasound for an explosion in the atmosphere. However, it is becoming clear
that there are many synergies between the waveform technologies that may be effectively exploited
in the future. The IDC is also required by the Treaty to routinely screen-out events considered
to be consistent with natural phenomena or non nuclear, man-made phenomena.
As noted above, analysis of the data from the waveform technologies alone does not uniquely
identify a source as nuclear. The smoking gun may come from data provided by the radionuclide
stations, an OSI, or even from National Technical Means, should a state choose to share such
means to help verify a Treaty violation.
In this article, we focus on the methods available to the forensic seismologist for source
identification. We discuss examples of special analysis of the so-called seismic problem events
(Figure 1), which some consider inevitable under a CTBT (Sykes 2002). We also discuss announced underground nuclear tests that have occurred since the Treaty was opened for signature
in 1996, including that by the Democratic People’s Republic of Korea (DPRK) on October 9,
2006 (and preceding false alarms).
We show that, in practice, the forensic seismologist often utilizes not only data from the IMS
network of stations and products from the IDC (Douglas et al. 1999, Bowers et al. 2001), but also
data from non-IMS networks of stations (Richards & Kim 1997, Hartse 1998, Bowers 2002, Kim
& Richards 2007). Non-IMS networks may be other international, regional, and national seismometer networks originally deployed for non-CTBT verification purposes, such as earthquake
hazard assessment, or for basic seismological research (van der Vink & Park 1994, Richards & Kim
1997, Kim & Richards 2007). There is now wide access to data from numerous seismic stations
via the Internet (Kradolfer 2000).
SOURCE IDENTIFICATION METHODS
There have been several substantial reviews of research into the problem of discriminating between earthquakes and underground explosions (Blandford 1977; Douglas 1981, 2007). Early
www.annualreviews.org • Forensic Seismology and the CTBT
211
ANRV374-EA37-10
ARI
27 March 2009
14:57
Kursk
August 12, 2008
60°N
Lop Nor
March 13, 2003
Annu. Rev. Earth Planet. Sci. 2009.37:209-236. Downloaded from arjournals.annualreviews.org
by University of British Columbia Library on 12/29/09. For personal use only.
North Korea
October 9, 2006
Pakistan
May 1998
30°N
0°
0° 30°E
India
May 1998
60°E
90°E
120°E
150°E
Figure 1
Locations of some seismic disturbances that have been of interest to forensic seismologists over the past
decade.
UNE: underground
nuclear explosion
mb : body-wave
magnitude (based on
the P wave amplitude
and period)
Ms : surface-wave
magnitude (based on
the Rayleigh wave
amplitude and period)
212
work concentrated on seismic methods using signals recorded at regional distances in the western
United States, defined as distances <1700 km (National Academy of Sciences 2002). Initial results
were not encouraging. However, clear P signals were seen by an experimental seismometer array
deployed in Wyoming, United States, from presumed underground nuclear explosions (UNEs)
in Kazakhstan and Algeria, leading to the realization that signals detected at large distances, i.e.,
3000–10,000 km, from the source (known as teleseismic signals) could be used for discrimination (Richards & Zavales 1996, Douglas 2007). The main teleseismic discriminants that emerged
from subsequent research are source depth, P complexity (simplicity), and the ratio of body- to
surface-wave magnitude (mb :Ms ).
For teleseismic discriminants, the identification threshold is usually taken as approximately
half a magnitude unit above the event-detection threshold. The threshold for detection by at
least three stations in the IMS primary seismic network is <3.5 mb for the vast majority of the
continental land mass 90% of the time (National Academy of Sciences 2002).
Since the late 1970s, there has been concerted research into discriminants that are effective
using signals recorded at regional distances, especially by the United States (Blandford 1981).
Research has shown that discriminants based on the spectral amplitude ratio of a single regional
phase can be ineffective (Blandford 1981, 1996), but that at high frequencies (>2 Hz) P/S amplitude
ratios are a promising discriminant (Blandford 1996).
The shift from teleseismic back to regional discrimination appears to have been driven by the
requirement to detect seismic signals from, and identify, UNEs with mb < 4. Signals from at least
three stations are usually required to be associated to form an event and estimate its location. Given
the distribution of seismic stations in the IMS network, it is likely that, for a source located on the
continental land mass, at least one station at regional distances will have clear signals. Thus, it has
been argued that there may be little difference in the event-detection and -identification thresholds
Bowers
·
Selby
Annu. Rev. Earth Planet. Sci. 2009.37:209-236. Downloaded from arjournals.annualreviews.org
by University of British Columbia Library on 12/29/09. For personal use only.
ANRV374-EA37-10
ARI
27 March 2009
14:57
for regional discriminants (National Academy of Sciences 2002), providing discriminants applied
at a single station at regional distances can be considered reliable.
Further, non-IMS seismic networks allow the event-detection threshold to be lowered below
3 mb for many parts of the world (National Academy of Sciences 2002) and increase the chance
of applying regional discriminants averaged over more than one station. It is estimated that,
globally, on average, there are more than 60,000 earthquakes a year with mb ≥ 3, compared with
approximately 7000 with mb ≥ 4. Lowering the detection threshold brings additional challenges,
not only caused by the significant increase in natural earthquakes that need to be considered, but
also because it is estimated that there are a few hundred mining explosions a year with mb ≥ 3
(National Academy of Sciences 2002).
In addition to earthquakes and large mining explosions, mine tremors appear explosion-like
on many traditional discriminants (Blandford 1996). Research suggests that such problem events
may be resolved using a combination of the first motion of teleseismic P (Bowers & Douglas
1997), combined body- and surface-wave modeling (Bowers & Walter 2002), and regional P/Lg
amplitude ratios (Bennett & McLaughlin 1997).
The IDC currently operates an experimental event screening process (identification of events
considered consistent with natural phenomena or nonnuclear, man-made phenomena) for all
events in the analyst Reviewed Event Bulletin (REB) with network-averaged mb ≥ 3.5 (Fisk
et al. 2002). The REB is produced by interactive analyst review of an automated event bulletin.
The experimental process has provisional criteria for screening using mb :Ms , depth, location
(onshore/offshore, with a minimum water depth constraint and nonobservation of unblocked underwater explosion-like hydroacoustic signals), and regional high-frequency P/S amplitude ratios.
There are more than 20,000 earthquakes each year with mb ≥ 3.5, so the global event screening
process adopted at the IDC will probably need to be highly automated to be cost-effective.
A key challenge for effective routine event screening is to define the explosion population under
a CTBT. The vast majority of presumed UNEs conducted have been fired at a depth sufficient
for containment of radioactivity, with little consideration of whether there is a permanent surface
expression or seismic waves are detected; detection via seismic signals or surface disturbance is
likely to be considered by a potential violator of the CTBT. Even for the well-instrumented tests,
it is not fully understood how underground explosions generate seismic waves (Douglas 2007).
Further, as Douglas (2007) states,
REB: Reviewed Event
Bulletin
TNT: trinitrotoluene
[t]he lack of understanding of the explosion source adds to the difficulty of setting screening criteria.
The available recordings from explosions are not some random sample from a population of possible
signals. Most of the explosions have been fired at a small number of test sites, so statistical comparisons
of earthquake and explosion signals may be unreliable.
However, it may be that not only do large tamped chemical and equivalent-yield UNEs produce similar seismic signals, but they also appear to follow similar scaling laws (Denny & Johnson
1991). Thus, experiments using small chemical explosions [a few tons equivalent or less of trinitrotoluene (TNT)] may help inform us about how larger-yield UNEs generate seismic waves.
One disadvantage of small-scale experiments is that gravity does not scale, so containment rules
(overburden pressure is usually assumed to be proportional to the cube-root of yield) and spall
effects will not necessarily be accurately reproduced.
For the forensic seismologist, an attractive alternate to the event screening strategy is to routinely attempt to detect statistical outliers to the earthquake population using promising discriminants (Fisk et al. 1996). Resources can then be targeted at identifying outliers in regions of interest
through special studies. Regularized discrimination analysis can be applied to combine candidate
www.annualreviews.org • Forensic Seismology and the CTBT
213
ANRV374-EA37-10
ARI
27 March 2009
14:57
discriminants, allowing for the level of explosion calibration information available, i.e., the transition from outlier detection to classical discrimination (Anderson & Taylor 2002, Anderson et al.
2007).
Some recent advances and remaining challenges for three of the experimental screening criteria
and candidate outlier detection discriminants are described below. The basic concepts and physics
on which these criteria are based have been used, along with waveform modeling and inversion
techniques, to resolve many problem events in special studies. The remainder of this article
gives examples of special studies to demonstrate the variety of methods available to the forensic
seismologist helping to verify compliance with the CTBT.
Annu. Rev. Earth Planet. Sci. 2009.37:209-236. Downloaded from arjournals.annualreviews.org
by University of British Columbia Library on 12/29/09. For personal use only.
Depth
The depth criterion is based on the idea that there is a physical limit to the depth at which a potential
violator could place a nuclear explosive device. For example, if the estimated source depth is
>10 km with 95% certainty, then the source is likely to be natural (Blandford 1977).
For depth estimates using only teleseismic P onset times, the estimated depth tends to trade off
with the origin time. When times from regional distance stations are available, the depth estimates
are also typically unreliable for depths less than 50 km and tend to have large uncertainty (Fisk
et al. 2002).
Approximately 15% of REB events are screened out using the experimental depth screening
criterion (Fisk et al. 2002). The experimental depth criterion is nominally a one-sided test that the
depth is >10 km at the 97.5% confidence interval. The current implementation has a constant
factor (to account for model uncertainty, e.g., +20 km), which is expected to reduce over time as
better models are developed (Fisk et al. 2002).
Depth phases, such as free-surface reflections pP and sP, should provide precise and reliable
depth estimates. However, reliable routine identification has turned out to be harder than expected.
Nonetheless, when candidate depth phases can be identified at three or more teleseismic stations,
then the amplitude ratios pP/P and sP/P have been widely used to assess whether these are consistent
with an earthquake model (double couple source) at a given depth (Pearce 1980, Douglas et al.
1999).
If the amplitude ratios are consistent with the earthquake model, then further confidence can
be gained in this interpretation by forward modeling (Douglas et al. 1972, Douglas 2007). Care
must be taken to ensure that an appropriate source-region model is used. The growing availability
of reliable geophysical models can further increase confidence in the modeling results (Bowers
et al. 2000). The most promising advances probably lie in the development of semiautomated
tools to help guide the analyst to routinely pick candidate depth phases (Murphy & Barker 2006,
Heyburn & Bowers 2008), the consistency of which can be assessed by the application of a set of
validation rules using regional models (Heyburn & Bowers 2008).
Observations of the near-regional phase Rg at frequencies approximately 0.5–1.0 Hz are diagnostic of a shallow source (depth less than approximately 3 km) (Bowers 1997, Myers et al.
1999, Kim & Richards 2007). However, 0.5 Hz Rg tends to be highly attenuated by scattering and
nonelastic attenuation, and is rarely observed at distances greater than 150 km. Indeed, there is
some evidence that Rg scattering close to the source is part of the explanation of Lg generation by
underground explosions (Blandford 1996, Myers et al. 1999).
The amplitude spectra of long-period surface waves (0.01–0.14 Hz) recorded at regional (and
teleseismic) distances can have significant depth resolution if the earthquake mechanism and depth
results in a notch in the Rayleigh spectrum (Douglas et al. 1971b, Patton 1998). Further research is
required to investigate the origin of Rayleigh spectral notches from underground explosions before
214
Bowers
·
Selby
ANRV374-EA37-10
ARI
27 March 2009
14:57
consideration as a routine screening method. However, analysis of surface-wave amplitude spectra
from a network of regional stations with good azimuthal coverage can increase confidence in the
resolution of problem events such as the March 13, 2003 Lop Nor disturbance (see Earthquake
Near Lop Nor, March 13, 2003, below).
Annu. Rev. Earth Planet. Sci. 2009.37:209-236. Downloaded from arjournals.annualreviews.org
by University of British Columbia Library on 12/29/09. For personal use only.
mb :Ms
It became clear from observations made in the 1960s that for a given distance-corrected shortperiod (∼1 s) P wave amplitude (mb ), the distance-corrected long-period (∼20 s) Rayleigh-wave
amplitude (Ms ) from presumed UNEs is generally smaller than from earthquakes. These observations formed the basis of the mb :Ms discriminant, although it was recognized early on that there
are some earthquakes that generate weak Rayleigh waves and so appear anomalous, or explosionlike, on mb :Ms (Douglas 1981). Network-averaged mb and Ms for discrimination and screening
need to be calculated assuming a shallow source (applying the amplitude-distance curves for zero
depth), as mb and Ms are source-depth dependent. The network-averaged mb and Ms also need to
be corrected for network data censoring effects. This can be achieved using maximum-likelihood
techniques (Lilwall 1987, Zaslavsky-Paltiel & Steinberg 2008).
Station and path corrections can be determined and applied to station mb and Ms to reduce individual magnitude bias and uncertainty in the network average (Marshall et al. 1979).
However, when deriving corrections to station magnitudes, care must again be taken to account for
data censoring (Lilwall 1987, Zaslavsky-Paltiel & Steinberg 2008) and identify possible trade-offs
between static corrections and the effects of nonisotropic radiation, especially if using earthquakes
for calibration (Bowers & Douglas 1998).
During development of the mb :Ms experimental screening criterion, it was recognized that there
is potential bias between the automatically measured network-averaged mb and Ms at the IDC,
and the respective magnitudes measured by an analyst from short- and long-period seismograms
as reported in global seismic event bulletins such as that by the National Earthquake Information
Center (NEIC). An initial study of magnitudes from earthquakes showed that the IDC networkaveraged mb and Ms was, on average, 0.4 and 0.1 units, respectively, less than those reported by the
NEIC (Fisk et al. 2002). Because the NEIC has mb and Ms observations from a large number of
presumed UNEs, the average NEIC/IDC magnitude biases from earthquakes were used to infer
the equivalent IDC explosion mb :Ms and hence define the experimental screening line (Fisk et al.
2002).
However, subsequent studies have shown that there is negligible difference between NEIC
and IDC network-averaged mb from presumed UNEs (Granville et al. 2002, Bowers et al. 2002).
The offset in NEIC and IDC network-averaged mb from earthquakes with NEIC mb ≥ 5 can
be explained by the different procedures used by the respective centers and the tendency for the
depth assigned to a common event by the NEIC to be greater than that assigned by the IDC
(Granville et al. 2005).
There is also a clear magnitude dependence to the IDC/NEIC network-averaged mb and Ms
bias. This can be attributed, at least in part, to data censoring effects of the different IDC and NEIC
station networks (Stevens & McLaughlin 2001, Zaslavsky-Paltiel & Steinberg 2008). Simulation
can provide insight into the network-dependent threshold above which data censoring can be
considered negligible. Preliminary results from simulation of network-averaged mb , suggest that
the threshold is approximately 4.5 mb and 5.5 mb for the (incomplete) IMS primary and NEIC-type
networks, respectively.
Ms for discrimination was originally defined based on Rayleigh amplitudes at periods around
20 s (Douglas 2007). Ms at the IDC is defined in a similar way (Rezapour & Pearce 1998) and
www.annualreviews.org • Forensic Seismology and the CTBT
P wave: primary
seismic body-wave
NEIC: National
Earthquake
Information Center
215
ANRV374-EA37-10
ARI
27 March 2009
14:57
is utilized in the experimental screening process (Stevens & McLaughlin 2001, Fisk et al. 2002).
Rayleigh waves from small underground explosions (with mb ∼ 4.5) are weak and are likely to
go undetected unless stations are at regional distances (the IMS auxiliary network and non-IMS
networks have proved to be valuable for detecting surface waves from problem events, as we
demonstrate later).
For continental propagation at regional distances, the largest amplitude is often the Airy phase,
with a dominant period significantly less than 20 s. To allow analysts to report Ms from seismograms
from small disturbances recorded at regional and teleseismic distances, and from more dispersed
waveforms following oceanic and mixed paths, a variable-period path correction was developed
(Marshall & Basham 1972).
Recently, a variable-period Ms measurement has been proposed based on bandpass Butterworth
filters applied in the time domain (Russell 2006, Bonner et al. 2006). Such a method appears to be
easily operationally implemented and can be regionalized (Russell 2006). A preliminary inversion
of global Ms measurements reported in the REB, for empirical station terms and two-dimensional
(2D) path corrections, shows a strong correlation between the path correction and distinct known
tectonic regions (Selby et al. 2003).
One of the challenges of Ms measurement is that surface waves from large earthquakes can
have durations of many hours and can potentially obscure surface waves from seismic disturbances
of interest (Douglas 2007). For example, the surface waves from the Pakistani announced UNE of
May 30, 1998 were obscured by those from an earthquake in Afghanistan approximately 30 min
earlier (Barker et al. 1998). Another problem is association, especially if the surface wave processing
is automated, as in the experimental IDC process. Robust methods of measuring Rayleigh wave
back azimuth have been shown to increase confidence in association (Selby 2001).
A problem that may fundamentally limit the usefulness of mb :Ms for event screening is that
the explosion and earthquake populations may converge for mb ≤ 4.5. This convergence has
been observed (Douglas 1981) and modeled (Stevens & Day 1985). However, there are workers who observe divergence for the earthquake and explosion populations in the Nevada Test
Site (NTS) region (Bonner et al. 2006). New data from the DPRK underground explosion presented later in this article may help throw some more light on this subject for the Eurasian
region.
Teleseismic P amplitudes are highly sensitive to attenuation in the mantle (Marshall et al. 1979),
and signals with small amplitudes often show complex P wave forms (Douglas 2007). However, the
scaling of observations of network-averaged mb with yield from announced UNEs are explained
to the first order by the Mueller-Murphy model of the UNE seismic source function (Mueller &
Murphy 1971, Murphy 1996). The coupling in the Mueller-Murphy model is dependent on the
source medium and depth (unlikely to be well known to the investigating forensic seismologist).
Furthermore, a regression analysis of data from underground explosions concluded that only
the low-frequency level and corner frequency of the underground explosion source amplitude
spectrum could be constrained (Denny & Johnson 1991). So it seems the underground explosion
source function for seismic waves is not yet well understood (Douglas 2007).
A further concern is the effect on mb :Ms of what is commonly referred to as tectonic release—
observations of long-period horizontally polarized S and surface waves not predicted by simple
elastic underground explosion models (Toksöz & Kehrer 1972). These anomalous observations
include Love waves and Rayleigh waves with polarity reversed relative to that expected from an
underground point elastic–explosion source (Toksöz & Kehrer 1972, Rygg 1979). The anomalous
observations are often interpreted by assuming a point elastic explosion with an additional doublecouple source (Ekström & Richards 1994). However, this interpretation is nonunique (Stevens &
Murphy 2002).
Annu. Rev. Earth Planet. Sci. 2009.37:209-236. Downloaded from arjournals.annualreviews.org
by University of British Columbia Library on 12/29/09. For personal use only.
NTS: Nevada Test
Site
216
Bowers
·
Selby
Annu. Rev. Earth Planet. Sci. 2009.37:209-236. Downloaded from arjournals.annualreviews.org
by University of British Columbia Library on 12/29/09. For personal use only.
ANRV374-EA37-10
ARI
27 March 2009
14:57
Empirically, presumed UNEs that show anomalous surface waves that have been studied
(Toksöz & Kehrer 1972, Ekström & Richards 1994, Fisk et al. 2002) do separate from the earthquake population on mb :Ms . However, the modeling study by Stevens & Day (1985), suggesting
convergence of the explosion and earthquake mb :Ms populations around mb ∼ 4.5, does not include
tectonic release in the explosion source model.
Physical mechanisms proposed to explain the anomalous surface wave observations from presumed UNEs are mainly related to nonlinear effects in the zone immediately surrounding the
explosion. Such mechanisms include (a) nonspherical effects due to heterogeneous and anisotropic
material, (b) cracking and block movement along preexisting lines of weakness, and (c) tectonic
release by the relaxation of preexisting stresses in the nonelastic zone surrounding the explosion
(Patton 1991). Triggering of earthquakes on faults in the vicinity of the explosion has also been
proposed.
It may be that at least some of the differences reported in the mb :Ms explosion populations for
Eurasia and NTS (Marshall & Basham 1972, Bonner et al. 2006) may be explained by differences
in the physical causes of the observations often attributed to tectonic release. Until the explosion
source model and physical causes of anomalous surface waves are understood and quantified, the
theoretical basis of mb :Ms for underground explosions remains unresolved. In later sections, we
show anomalous surface waves from the announced UNEs detonated in India and Pakistan on
May 11 and 28, 1998, respectively.
Regional P/S Amplitude Ratios
The spatial distribution of IMS and non-IMS seismic stations means it is likely that, for a source
located on the continental land mass, at least one station at regional distances will have clear
signals. If the aim is to identify UNEs with mb < 3.5, then methods utilizing seismic signals
recorded at regional distances provide a valuable tool for the forensic seismologist. The most
promising methods utilize P/S amplitude ratios.
Empirical observations. Research since the early 1980s suggests that the regional P/S amplitude
ratio at high frequencies ( f > 2 Hz) shows promise as a discriminant (Walter et al. 1995, Blandford
1996, Fisk 2006) but is highly region-dependent (due to crustal and upper mantle structure).
Usually the amplitude ratio is measured from P and S recorded on the vertical component only,
although some studies have shown that three-component amplitude measurements can improve
discrimination (Kim et al. 1997). If there is strong scattering, then it is unlikely to matter from
which component the amplitude is measured, and the P/S ratio is stable (Blandford 1996) as the
P and S source-radiation patterns are obscured (Zhang et al. 2002). However, if there is weak
scattering, then three-component measurements may be able to exploit differences in the P and
polarized-S radiation patterns from earthquake and explosion sources (Lilwall 1988, Bowers et al.
2001, Bowers 2002).
There appears to be some variation in the definition of regional distance, !. For example,
the National Academy of Sciences (2002) prefers ! ≤ 15◦ , whereas the experimental regional
screening process at the IDC uses P and S recorded at stations with 3◦ ≤ ! ≤ 17◦ (Bottone et al.
2002).
Studies show that there can be large variations in the amplitudes observed at sensors separated
by less than 100 km. At source-receiver distances of approximately 20◦ , P amplitudes in the
1–3 Hz passband, from a presumed UNE and an earthquake, vary by a factor of up to eight
across the NORSAR array in Norway, whereas S amplitudes appear to vary little (Ringdal et al.
2001). At a source-receiver distance of approximately 10◦ , P amplitudes in the 0.2–8.0 Hz passband
www.annualreviews.org • Forensic Seismology and the CTBT
217
ANRV374-EA37-10
ARI
27 March 2009
14:57
recorded by the Kyrgyzstan network vary by a factor of 20 (Xie & Patton 1999). Both observations
of variations in P amplitudes have been attributed to strong three-dimensional (3D) structure in the
vicinity of the recording network. Possible variations in P and S amplitudes need to be considered
if data from surrogate stations are used to help calibrate new IMS stations. For example, the
seismic station MAKZ in east Kazakhstan (calibrated by Bottone et al. 2002) is 24 km from the
IMS station MKAR used by the IDC experimental regional screening process.
Annu. Rev. Earth Planet. Sci. 2009.37:209-236. Downloaded from arjournals.annualreviews.org
by University of British Columbia Library on 12/29/09. For personal use only.
Path corrections and transportability. To account for the regional variations in propagation
of P and S, various schemes have been proposed to determine empirical distance-dependent
corrections. It was found that using optimal spatial prediction [often referred to as kriging (Schultz
et al. 1998)] based on the observed P/S ratios in common fixed passbands significantly improved
discrimination (Rodgers et al. 1999b)
The experimental regional screening process at the IDC uses a fixed passband 6–8 Hz and
Pn /Smax ratios, where Smax is either Sn or Lg , with a minimum signal-to-noise criterion for each
phase (Bottone et al. 2002). The kriging algorithm used to derive the station-specific 2D correction
surfaces for Pn /Sn and Pn /Lg ratios used by the IDC process converges to the global mean value (for
earthquakes) when no calibration data are available and to the local mean when there are sufficient
calibration data. Intermediate values depend on an exponential distance-damping function defined
by a correlation length estimated to be 5–7◦ from the data by semivariogram modeling (Bottone
et al. 2002).
The associated uncertainty surfaces are based on a calibration variance and the variance from
the earthquake data residuals, both estimated by semivariogram modeling (Bottone et al. 2002).
To calculate the experimental screening score (criterion), the explosion P/S variance also needs
to be set. This can be estimated from an UNE dataset by semivariogram modeling, but we must
consider the possibility that the available presumed UNE dataset may not be drawn from the same
population as UNEs under a CTBT. If P/S is available from more than one station, then the scores
can be averaged (Bottone et al. 2002).
Alternatively, various P/S ratios can be formed from the regional P phases Pn , Pg , and S
phases Sn , Lg , either in common or different (cross-spectral) passbands for P and S. It may be
that for a particular region a certain combination offers the most promising discriminant (Walter
et al. 1995, Fan et al. 2002, Rodgers & Walter 2002), or a number of combinations are to be
considered as part of an outlier detection strategy (Fisk et al. 1996).
A model-based approach with source (earthquake model) and distance (path) frequencydependent corrections has been developed for each regional phase, Pn , Pg , Sn , and Lg (Taylor
et al. 2002, Walter & Taylor 2001), for a single station. This approach allows the whole spectrum
of the source and path effects to be modeled for each regional phase, allowing flexibility. The
model constrains the regional mean, upon which the calibration data can be kriged. Further, the
model-based approach allows the mean of the path effects for each phase to be predicted from
geophysical experience in poorly or uncalibrated regions (Taylor et al. 2002). This predictive potential is attractive, as initially poor and uncalibrated regions are likely to be the norm for global
monitoring under a CTBT. Over time, models should improve (e.g., resolving the trade-off between geometrical spreading and nonelastic attenuation), and the number of calibration events
could increase, either through confidence-building measures or through serendipity.
Several workers have reported improved discrimination if the P/S ratios are averaged over a
network (Walter et al. 1995, Blandford 1996, Xie & Patton 1999). It has also been shown that
network-averaged Pn , Sn , and Lg relative spectra from pairs of nearby explosions and earthquakes
(so path effects cancel) at Lop Nor, China, with different magnitudes, are easier to interpret than at
a single station (Fisk 2006). Network-averaged P/S ratios might be preferred given the potential
218
Bowers
·
Selby
ANRV374-EA37-10
ARI
27 March 2009
14:57
variation in P amplitude described above. Significant P amplitude variation may confound the
screening, discrimination, or outlier strategies if not represented by the correction and uncertainty
surfaces.
Another challenge is to correctly associate P and S phases with the event under consideration.
There is at least one example of an automated process resulting in misassociation of P and S
with the May 28, 1998, Pakistani announced UNE and apparent subsequent failure of the P/S
discriminant ( Jenkins & Sereno 2001, Bowers et al. 2002).
S wave: secondary or
shear seismic
body-wave
Annu. Rev. Earth Planet. Sci. 2009.37:209-236. Downloaded from arjournals.annualreviews.org
by University of British Columbia Library on 12/29/09. For personal use only.
Research into how explosions generate S waves. Over the past decade or so there has been a
concerted effort, especially by United States researchers, to attempt to understand how explosions
generate high-frequency S waves. The main candidate mechanisms proposed are:
1. interaction of curved P wavefronts with the free-surface above the source (Lilwall 1988)
and regional phase propagation in realistic 3D earth structure with topography (Myers
et al. 2003);
2. scattering from Rg -to-S. This may be significant in the 1–5 Hz passband and depends on
the explosion source depth (Myers et al. 1999);
3. significant deviations from a point explosion source in an elastic medium with a free surface,
such as the effect of spall (Blandford 1996, Stevens et al. 2006);
4. cracking in some volume surrounding the detonation point (Blandford 1996).
There are an increasing number of studies that suggest the corner frequency of regional Sn and
Lg from underground explosions is significantly less than the corresponding P corner frequency
(Xie & Patton 1999, Fisk 2006, 2007). Differing P and S corner frequencies would suggest that
P-to-S conversion is not the dominant mechanism for S generation from underground explosions,
as such a mechanism would be expected to produce S spectra with the same corner frequency as
P (Fisk 2006). Further, recent observations and modeling suggest Rg -to-S scattering may be less
important than previously thought (Stevens et al. 2006).
Fisk (2006) conjectured that the S-wave spectrum from underground explosions in hard rock
can be represented by the Mueller-Murphy explosion P wave source spectrum (Mueller & Murphy
1971), with S-wave corner frequency lower than P by the ratio of P to S wave speeds in the
material around the source. Combining the Fisk conjecture for the explosion source with standard
earthquake source models seems to explain why P/S discriminants perform well at frequencies
above approximately 2 Hz but not at lower frequencies (Fisk 2006).
One implication of the Fisk conjecture is that the S waves are generated either at or in the
immediate vicinity of the source (Fisk 2006). The conjecture also predicts that P/S criteria should
only be reliable at frequencies around and above the P corner frequency, i.e., a magnitude and
explosion source-depth dependence. Yet another consequence of the conjecture is that such scaling
of S corner frequencies from explosions is unlikely to be consistent with nonlinear spall effects,
given the scaling appears to hold for differing emplacement media (e.g., tunnel versus shaft) and
firing depths (Fisk 2006).
The physical cause of S waves from underground explosions remains poorly understood.
Current research efforts seem to be concentrating on the following:
1. attempting to quantify the uncertainty in and resolving the trade-off between parameters
describing the path and station effects (required to estimate the explosion source function);
2. confirming the observations that suggest the underground explosion S corner frequency is
less than and proportional to that of P;
3. explaining the observations of high-frequency S, Lg , and surface waves (often showing
evidence of tectonic release) from presumed UNEs.
www.annualreviews.org • Forensic Seismology and the CTBT
219
ANRV374-EA37-10
ARI
27 March 2009
14:57
INDIAN UNDERGROUND TEST OF MAY 11, 1998
UTC: Universal
Coordinated Time
Annu. Rev. Earth Planet. Sci. 2009.37:209-236. Downloaded from arjournals.annualreviews.org
by University of British Columbia Library on 12/29/09. For personal use only.
kt: equivalent energy
released by a kiloton of
TNT (∼4.2 × 1012 J)
A seismic disturbance, located at 27.081◦ N, 71.738◦ E with origin time 10:13:44.2 UTC (subsequently referred to as IND 980511), is directly associated with the announcement by India of
the simultaneous detonation of three devices with nuclear yields of 43, 12, and <1 kilotons (kt)
(Barker et al. 1998).
High frequency P/S ratios (>2 Hz) for IND 980511 indicate that the disturbance is an outlier
to the earthquake population defined by common regional phase-ratio discriminants, such as
Pn /Lg and Pn /Sn . Interestingly, these discriminants also seem effective at frequencies as low as 0.5–
2.0 Hz (Rodgers & Walter 2002).
Surface waves generated by IND 980511 were observed at many stations across Eurasia. Unfortunately, Rayleigh waves reported in the NEIC bulletin were subsequently shown to be misassociated (i.e., they originated from a different seismic disturbance); this misled some authors
(Evernden 1998, Sikka et al. 2001). Using carefully associated Rayleigh waves, the networkaveraged Ms for IND 980511 is 3.32 (Douglas et al. 2002).
The prototype IDC REB reported a network-averaged mb of 5.0, but did not report any
associated Ms measurements. Using a network-averaged Ms = 3.3, IND 980511 is an outlier
to the earthquake population on an mb :Ms plot (Figure 2) and is not screened out using the
experimental mb :Ms criteria (Fisk et al. 2002). Thus, the seismic waves generated by IND 980511
are consistent with an explosion source. However, modeling of the surface waves recorded at nonIMS stations NIL (Pakistan) and HYB (India) showed that the Rayleigh waves observed had the
opposite polarity from that expected from a simple elastic explosion source (Rodgers et al. 1999a).
Subsequent work demonstrated that Rayleigh waves were reversed at all stations where they were
observed (Selby et al. 2004), with the best-fit model having reversals at all azimuths.
The reversed polarity of Rayleigh waves from presumed UNEs was briefly described in the
section entitled mb :Ms , and it has been observed at NTS (Toksöz & Kehrer 1972) and at the
Shagan River test site, eastern Kazakhstan (Rygg 1979, Herrin & Goforth 1986). The reversals
are believed to be due to the interference of the Rayleigh waves from the explosion source, with
the Rayleigh waves due to the accompanying tectonic release usually modeled as a double couple.
Ekström & Richards (1994) calculate Rayleigh and Love radiation patterns from 71 presumed
UNEs at Shagan River. Only four of these explosions show Rayleigh reversal at all azimuths
(the smallest having 5.29 mb ); a further three have a Rayleigh radiation pattern dominated by
reversals.
If Rayleigh waves generated by tectonic release can be larger than those from the point explosion
source, it is not clear why the mb :Ms criterion should work for explosions of any size. Toksöz &
Kehrer (1972) consider this issue and conclude that tectonic release does not affect the networkaveraged Ms if good azimuthal coverage is available. However, Toksöz & Kehrer (1972) constrain
tectonic release at NTS to be a strike-slip mechanism, and their argument does not appear to
apply where the tectonic release is not believed to be strike-slip—e.g., Shagan River (Ekström
& Richards 1994) and IND 980511. To conclude, the study of long-period waveforms from
IND 980511 confirms that the mechanism for the generation of surface waves by underground
explosions is not understood, and that this has implications for the applicability of the mb :Ms
criterion.
PAKISTANI UNDERGROUND TEST OF MAY 28, 1998
Pakistan announced that it had conducted an underground nuclear test involving five explosive
devices on May 28, 1998 (subsequently referred to as PAK 980528). The location is reported as
220
Bowers
·
Selby
ANRV374-EA37-10
ARI
27 March 2009
14:57
ID
Ce M
xp s =
er 1
im .25
en m
ta b –
ls 2
cre .2
en
ing
l in
e
7
Surface wave magnitude (Ms)
Annu. Rev. Earth Planet. Sci. 2009.37:209-236. Downloaded from arjournals.annualreviews.org
by University of British Columbia Library on 12/29/09. For personal use only.
6
REB Eurasian events 2006
5
China 950817
Lop Nor 030313
China 960608
4
Pakistan 980528
India 980511
3
China 960729
Presumed earthquakes
DPRK 061009
2
Presumed explosions
2
3
4
5
6
7
Body wave magnitude (mb)
Figure 2
mb :Ms for recent Eurasian underground explosions and earthquakes. Magnitudes for earthquakes shown are from the Reviewed Event
Bulletin (REB) except for Lop Nor 030313 (Selby et al. 2005). Magnitudes for the explosion in North Korea (DPRK 061009) are from
Selby & Bowers (2007), and for the explosions in India 980511 and Pakistan 980528 are from Bowers et al. (2002), except Ms for India
980511, which is from Douglas et al. (2002). For the Chinese explosions, mb is from the REB and Ms is from Bonner et al. (2006).
28.792◦ N, 64.948◦ E from satellite imagery (Albright et al. 1999), with an origin time of 10:16:17.0
UTC estimated from seismic signals (Barker et al. 1998). The prototype IDC REB reported
network-averaged mb and Ms of 4.9 and 3.6, respectively. PAK 980528 was not screened out using
the experimental mb :Ms criteria (Fisk et al. 2002) and was not considered for the experimental P/S
screening (Bottone et al. 2002), presumably because data from NIL were not available. Nonetheless, PAK 980528 appears to be more earthquake-like than the vast majority of presumed UNEs
considered when the experimental mb :Ms screening criteria was developed.
As briefly mentioned earlier, Jenkins & Sereno (2001) failed to identify PAK 980528 as an
explosion using regional high-frequency P/S amplitude ratios. The apparent failure of the highfrequency P/S criteria seems to be due to the misassociation of P from a near-regional earthquake
to S from PAK 980528 (Bowers et al. 2002).
Bowers et al. (2002) estimated a network-averaged Ms equal to 3.44 from carefully associated
Rayleigh waves using a variable-period method with path-corrections (Marshall & Basham 1972),
noting that PAK 980528 falls between the historical presumed UNE and earthquake populations
for Eurasia (Figure 2). Thus, PAK 980528 may arouse suspicion using the mb :Ms criteria, but it
is also consistent with deep-lithospheric Eurasian earthquakes.
www.annualreviews.org • Forensic Seismology and the CTBT
221
ANRV374-EA37-10
ARI
27 March 2009
14:57
100
50
0
–50
RAYN Rayleigh
–100
3.5
100
Annu. Rev. Earth Planet. Sci. 2009.37:209-236. Downloaded from arjournals.annualreviews.org
by University of British Columbia Library on 12/29/09. For personal use only.
Displacement (nm)
50
0
–50
RAYN Love
–100
3.5
100
50
0
–50
–100
CMAR Rayleigh
3.5
CMAR Love
3.5
100
50
0
–50
–100
0
500
1000
1500
Time after 10:16:17 (s)
Figure 3
Surface-wave seismograms from the underground explosion in Pakistan on May 28, 1998 recorded at
stations RAYN (Saudi Arabia) and CMAR (Thailand). Seismograms are filtered in the passband 0.025–
0.045 Hz. At both stations, the Love waves (red ) are larger than the Rayleigh waves (blue). The vertical line
indicates the time of arrival for a group wave speed of 3.5 km/s.
Although P onset times from 63 stations were reported in the prototype IDC REB for PAK
980528, the vast majority of the teleseismic P waves recorded from PAK 980528 are unusually
complex compared with those typically observed from underground explosions. Further, there are
strong Love waves recorded at the non-IMS station in Saudi Arabia (RAYN) and an IMS station
in Thailand (CMAR) (Figure 3), indicating that there was significant tectonic release. However,
simple P seismograms were recorded by at least three IMS stations (in Australia, Antarctica, and
the United States). Bowers et al. (2002) showed that these three simple P seismograms are inconsistent with an earthquake source at a depth of at least 5 km, implying that the source is shallow
and thus should arouse suspicion using the mb :Ms criteria.
KURSK SUBMARINE DISASTER, AUGUST 12, 2000
Seismologists at the NORSAR group in Norway reported two nearly colocated seismic disturbances on August 12, 2000, with epicenters in the Barents Sea (Ringdal et al. 2000). Seismic
disturbances in the Barents and Kara Sea regions (Figure 1), which have a low level of natural
seismicity, are of interest due to the proximity of the Russian nuclear test site at Novaya Zemlya
(Koper et al. 2001). The first and second seismic disturbances have local magnitudes of 1.5 and
222
Bowers
·
Selby
ANRV374-EA37-10
ARI
27 March 2009
14:57
Mean ARCES amplitude spectra: Disturbance 2
P
Lg
Lg coda
Noise
Normalized spectral amplitude
Annu. Rev. Earth Planet. Sci. 2009.37:209-236. Downloaded from arjournals.annualreviews.org
by University of British Columbia Library on 12/29/09. For personal use only.
101
100
10–1
10–2
10–3
0
2
4
6
8
10
12
14
Frequency (Hz)
Figure 4
Array-averaged amplitude spectra from time windows corresponding to P, Lg , and Lg coda from the second
disturbance associated with the sinking of the Kursk submarine on August 12, 2000. The amplitude
spectrum of the noise preceding the P onset of the signal from the second disturbance (i.e., the coda from the
first disturbance) is also shown.
3.5 respectively (Ringdal et al. 2000). P signals from the larger second disturbance were detected
as far away as Alaska and reported in the REB.
Seismic signals from the two disturbances were clearly recorded by the IMS seismometer array
(25 sensors) in northern Norway (ARCES). ARCES is approximately 470 km from the reported
location of the tragic sinking of the Kursk submarine at 69.6166◦ N, 37.5708◦ E (Schweitzer 2002).
The epicenter of the second disturbance estimated by NORSAR is within 6 km of the reported
location, and hence the seismic signals observed were associated with the sinking of the Kursk
(Ringdal et al. 2000).
The origin time of the second disturbance is estimated as 07:30:42.4 UTC; relative times
measured using cross-correlation show the first disturbance occurred 135.8 s earlier, located within
a few hundred meters of the second. Further, the relatively high cross-correlation coefficients for
P and S at ARCES have been interpreted as indicating similar, but not identical, source and
propagation characteristics (Schweitzer 2002).
Figure 4 shows the array-averaged amplitude spectra from signals recorded at ARCES from
the second disturbance. The P, Lg , and Lg coda show a common modulation with a regular interval frequency of approximately 1.45 Hz, and clear spectral notches at approximately 6
and 12 Hz. Given the very different propagation paths, the similarity of the modulation and
www.annualreviews.org • Forensic Seismology and the CTBT
223
ARI
27 March 2009
14:57
notches in the regional phase amplitude spectra suggests that these are related to the seismic
source.
The 1.45 Hz modulation frequency is diagnostic of the bubble pulse from an underwater
explosion (Koper et al. 2001, Sebe et al. 2005). Many decades of experience with underwater
chemical explosions suggest the yield can be estimated from the bubble-pulse frequency if the
source depth is known. It turns out that the notches in the ARCES amplitude spectra (Figure 4)
constrain the source depth.
Figure 5 shows notches in synthetic far-field amplitude spectra that are dependent on the
source depth in a water layer over a solid half-space. We can see that for a water layer 120 m
thick, synthetic spectral notches at approximately 6 and 12 Hz are seen if the source is on the sea
floor. Similar observations and modeling lead to the conclusion that the second Kursk explosion
occurred on the sea floor (Koper et al. 2001, Schweitzer 2002, Sebe et al. 2005). The depth/bubblepulse frequency and seismic magnitude give a yield estimate roughly equivalent to approximately
five tons of TNT (in the range of two to eight tons).
Similar analysis of the regional phase amplitude spectra of signals recorded at ARCES from the
first disturbance is not possible due to the poor signal-to-noise ratio. However, the similarity of the
signals from the first and second disturbances suggests that the source of the first disturbance is also
an explosion. Semblance analysis of the beamformed regional phases suggests the beamformed P
amplitude spectrum is reliable over the passband 2–10 Hz. The beamformed P spectrum shows
no evidence of a bubble pulse, or of a notch at approximately 6 Hz, which is consistent with the
first explosion occurring at a depth of approximately 80 m or less (Figure 5) and being completely
contained by the hull of the submarine. The estimate of the yield from seismic magnitude of the
first explosion is in the range of 2 to 80 kg of TNT. However, decoupling experiments underwater
show signals are muffled by a factor of between 5 and 10 (Khristoforov 1996). If the potential
muffling effect of the submarine’s hull is considered, the yield estimate of the first explosion is
revised to the range of 10 to 800 kg of TNT.
Several authors have referred to observations of negative first-motion of P from the second Kursk disturbance, which is more consistent with an implosion than an explosion source
(Koper et al. 2001, Schweitzer 2002). Schweitzer (2002) compares filtered (passband 1.5–8.0 Hz)
beamformed P onsets at IMS seismometer arrays ARCES and FINES (Finland) from the second
Kursk disturbance, with those from a local-magnitude 2.5 underwater explosion conducted by the
Russian navy. The P onsets on the filtered seismograms from the explosion are clearly impulsive with positive first motion, whereas those from the second Kursk disturbance are apparently
negative (Schweitzer 2002).
Figure 6 shows the beamformed ARCES and FINES seismograms from the second Kursk
disturbance, both unfiltered and optimally filtered to suppress sinusoidal noise (Douglas 1997).
It seems clear that the first motions are positive, and hence consistent with an explosive source.
Figure 6 also shows the FINES seismogram filtered to simulate a wide-band velocity seismograph. It becomes clear that the rise-time of the P-pulse is comparatively slow relative to an
impulsive onset, resulting in weak first motion in the short-period seismograph passband. The
forensic seismologist should be aware of the potential pitfalls of attempting to read first motion from bandpass filtered seismograms, or narrow-band seismograph systems (Douglas et al.
1997).
It has been known for decades that underwater explosions are remarkably efficient at generating
seismic signals at long range—a 10-ton chemical explosion fired in the North Sea in July 1971
generated teleseismic P waves detected in Brazil and Australia ( Jacob & Willmore 1972). Thus,
the data from the IMS seismic network can be used to detect and identify small explosions fired
underwater in areas where paths to the hydroacoustic network are blocked.
Annu. Rev. Earth Planet. Sci. 2009.37:209-236. Downloaded from arjournals.annualreviews.org
by University of British Columbia Library on 12/29/09. For personal use only.
ANRV374-EA37-10
224
Bowers
·
Selby
ANRV374-EA37-10
ARI
27 March 2009
14:57
a
h 0m
h 20m
Log(amplitude)
Annu. Rev. Earth Planet. Sci. 2009.37:209-236. Downloaded from arjournals.annualreviews.org
by University of British Columbia Library on 12/29/09. For personal use only.
h 40m
h 60m
h 80m
h 100m
h 120m
0
5
10
15
Frequency (Hz)
b
Free surface
VP = 1500 m s–1
Thickness = 120 m
Depth 80 m
VP = 8000 m s–1
pnwP
PnwP
Model seismogram
0.0
0.5
1.0
1.5
Time (s)
2.0
2.5
3.0
Figure 5
(a) Amplitude spectra for varying source depths in the model below. (b) Model and example synthetic seismogram for an impulsive
seismic source (at a depth of 80 m) in a water layer (120 m thick) over a half-space.
www.annualreviews.org • Forensic Seismology and the CTBT
225
ANRV374-EA37-10
ARI
27 March 2009
14:57
ARCES Pn beam for second KURSK disturbance (distance 470 km)
a
40
SP P Beam
20
0
Annu. Rev. Earth Planet. Sci. 2009.37:209-236. Downloaded from arjournals.annualreviews.org
by University of British Columbia Library on 12/29/09. For personal use only.
Displacement (nm at 1 hz)
–20
–40
6
4
2
0
–2
–4
–6
40
Filtered SP
Estimated noise
20
0
–20
–40
0
b
2
4
6
8
10
Time (s)
12
14
16
18
20
18
20
FINES Pn beam for second KURSK disturbance (distance 1050 km)
SP P Beam
5
Displacement (nm at 1 hz)
0
–5
5
Filtered SP
0
–5
Filtered VBB
5
0
–5
0
226
Bowers
·
Selby
2
4
6
8
10
Time (s)
12
14
16
ANRV374-EA37-10
ARI
27 March 2009
14:57
Annu. Rev. Earth Planet. Sci. 2009.37:209-236. Downloaded from arjournals.annualreviews.org
by University of British Columbia Library on 12/29/09. For personal use only.
EARTHQUAKE NEAR LOP NOR, MARCH 13, 2003
The REB reported a seismic disturbance on March 13, 2003 near the Chinese nuclear test site
at Lop Nor (epicenter 41.776◦ N, 89.078◦ E, origin time 15:07:08.3 UTC, 4.3 mb ), subsequently
referred to as Lop Nor 030313. The REB associated P waves from 21 IMS stations, but no
Ms was reported. The location of this seismic disturbance and the remarkably simple shortperiod teleseismic P waves, all with upward first motion (i.e., explosion-like), warranted further
investigation.
Selby et al. (2005) identify Lop Nor 030313 as an earthquake with mb :Ms (Figure 2) using
the variable-period Ms measurement, with path corrections (Marshall & Basham 1972). Ms was
measured from Rayleigh waves recorded by 12 stations, with network-averaged Ms of 3.63 (Selby
et al. 2005). Selby et al. (2005) were able to obtain data from 16 stations within 30◦ of the Lop Nor
030313 epicenter. Although most of these 16 stations are not IMS stations, they can be considered
substitutes for the completed IMS seismic network in the region (Selby et al. 2005). Modeling the
surface-wave (especially Rayleigh) amplitude spectra from these 16 stations, Selby et al. (2005)
demonstrate that Lop Nor 030313 is an earthquake with a 40◦ reverse-dip-slip mechanism at
a depth of approximately 6 km. This result is further supported by full waveform modeling of
three-component data from a non-IMS station WMQ, at a distance of approximately 250 km
(Selby et al. 2005).
Historically, mb :Ms has been a robust discriminant for disturbances with mb > 4.5 (although
perhaps not at smaller magnitudes). The cause of differences between the mb :Ms ratio for earthquakes and explosions has been discussed in several papers (Douglas et al. 1972, Stevens & Day
1985, Douglas 2007). Douglas et al. (1972) demonstrate that shallow earthquakes with near-45◦
reverse-dip-slip mechanisms might be expected to be anomalous on the mb :Ms criteria because
(a) P waves will show upward first motion at all teleseismic stations (as will P waves from UNEs),
(b) for source depths less than approximately 10 km, it is difficult to clearly identify the depth
phase pP (and for very shallow sources pP will constructively interfere with P, increasing mb ),
(c) Rayleigh wave amplitude spectra for this type of focal mechanism have a zero or null at a period
dependent on the source depth and azimuth to the recording station (Tsai & Aki 1970; Douglas
et al. 1971a,b), and this may reduce Ms .
The modeling of Rayleigh wave amplitude spectra has successfully been used to estimate the
focal mechanisms and depths of several earthquakes (in addition to Lop Nor 030313) in northwest
China (Fox et al. 2005). It appears that the method is especially applicable to shallow continental
dip-slip earthquakes, which may be difficult to identify using the mb :Ms criterion.
NORTH KOREA 2004–2006
Concerns about the development of nuclear weapons by North Korea (DPRK) were heightened
by its withdrawal from the Nuclear Non-Proliferation Treaty in April 2003. The possibility that
the DPRK might test a nuclear weapon led to increased interest in seismic monitoring of the
←−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−
Figure 6
P onsets from the second disturbance associated with the sinking of the Kursk submarine on August 12,
2000. (a) Comparison of the unfiltered short-period array beam at ARCES with the same trace filtered to
estimate the signal assuming the noise is a sinusoid (Douglas 1997), and the estimated noise. (b) Comparison
of the unfiltered short-period array beam at FINES with the signal estimated using the method of Douglas
(1997). The bottom trace is the filtered beam-signal estimate converted to an instrument response that is
proportional to velocity.
www.annualreviews.org • Forensic Seismology and the CTBT
227
ANRV374-EA37-10
ARI
27 March 2009
14:57
MDJ
44°N Earthquake
September 8, 2004 14:24:20.5 UT
41.984N 128.146E
Annu. Rev. Earth Planet. Sci. 2009.37:209-236. Downloaded from arjournals.annualreviews.org
by University of British Columbia Library on 12/29/09. For personal use only.
China
"Mushroom cloud"
September 9, 2004 02:00 UT
42°N Nuclear test
October 9, 2006 01:35:27.6 UT
41.253N 129.048E
Hydroelectric project
September 16, 2004
REB event
April 22, 2004 03:19:11.3 UT
39.819N 124.413E
40°N Elevation (m)
3000
1500
North Korea
0
–1500
–3000
38°N INCN
124°E 126°E –4500
South Korea
km
KS31
128°E 130°E 0
50
100
Figure 7
Locations of what turned out to be seismic nonevents and genuine seismic events of interest to the forensic seismologist in the North
Korean region over the past five years.
region. During 2004, two events occurred for which the application of forensic seismological
techniques was able to help clarify confusing press reports.
On April 22, 2004, a large explosion occurred near the railway station at Ryongchon in northwestern North Korea (see Figure 7). Little is known about the explosion, although the North
Korean news agency KCNA reported that it was an accident caused by an electrical contact during
the shunting of wagons loaded with ammonium nitrate fertilizer (KCNA 2004, BBC 2004a). Press
228
Bowers
·
Selby
Annu. Rev. Earth Planet. Sci. 2009.37:209-236. Downloaded from arjournals.annualreviews.org
by University of British Columbia Library on 12/29/09. For personal use only.
ANRV374-EA37-10
ARI
27 March 2009
14:57
reports (Chosun 2004a), apparently based on seismological findings, suggested that the size of the
explosion was much larger than reported by the DPRK government. However, analysis of the
data showed that no seismic signals could convincingly be demonstrated to originate from this
explosion.
On September 12, 2004 the media reported (BBC 2004b) that a mushroom cloud had been
observed in North Korea near the border with China (Figure 7). It was also reported that seismic
signals had been detected and may have been associated with the cause of the mushroom cloud.
Analysis of waveform data demonstrated that the only seismic event detected in the area concerned
was located near 42.0◦ N, 128.1◦ E (see Figure 7), which was more than 100 km from the supposed
mushroom cloud location. The DPRK claimed that the mushroom cloud was due to demolition
activity related to a hydroelectric project, which was at a third location (Figure 7). Subsequently,
consensus emerged that the mushroom cloud was likely a natural formation (Chosun 2004b).
Underground Test of October 9, 2006
The DPRK announced it had conducted an underground nuclear test on October 9, 2006 (KCNA
2006); the resulting seismic disturbance is subsequently referred to as DPRK 061009. This was the
first nuclear explosion announced since May 1998, ending a de facto global moratorium that had
lasted for more than eight years. The location and origin time of the explosion were well resolved
by teleseismic and regional data soon after the explosion occurred (CTBTO 2007), demonstrating
the ability of the IMS to detect explosions of mb ∼ 4. The United Kingdom National Data Center
(NDC) estimated the epicenter of the explosion as 41.253◦ N, 129.048◦ E, with an origin time of
01:35:27.6 UTC. DPRK 061009 has a maximum-likelihood mb = 3.94 and Ms = 2.83 (Selby
& Bowers 2007). Other agencies calculated similar values. DPRK 061009, an announced nuclear
explosion in a new location, provided the opportunity to test the power of forensic seismology to
identify an explosion source in a new geological context.
Depth. Unless local stations are available, the depth of a shallow seismic source is difficult to
estimate from the phase onset times because the depth trades off with origin time. However, there
are some indications that the source of DPRK 061009 was shallow. The clear Rg phase observed
at non-IMS station MDJ (China) suggests a shallow source; Rg is not observed at this station from
a nearby mid-crustal depth earthquake (Kim & Richards 2007). There are simple teleseismic P
seismograms at several stations (Figure 8), lacking evidence of clear depth phases pP and sP. These
seismic observations suggest that DPRK 061009 is either a shallow earthquake or an explosion.
Further, teleseismic P polarities together with the phase of Rayleigh waves observed at the closest
stations (Figure 8) are consistent with a shallow simple elastic explosion source model, although
they are not sufficient to formally rule out a shallow earthquake source.
Regional P/S ratios. Two studies (Kim & Richards 2007, Walter et al. 2007) report that DPRK
061009 is an outlier to the earthquake population defined using regional high frequency P/S
ratios. However, as described in the section entitled Regional P/S Amplitude Ratios, this criterion
is empirical, and its physical basis is not fully understood.
mb :Ms . Figure 2 shows mb versus Ms for several recent Eurasian presumed UNEs and Eurasian
earthquakes during 2006 from the IDC REB. Unlike the explosions in India, Pakistan, and China,
DPRK 061009 falls on the edge of the earthquake population from the REB. This leads to doubts
that the experimental screening criterion used by the IDC is applicable to seismic disturbances
with small magnitude. Currently, it is unclear whether DPRK 061009 is anomalous, or whether the
www.annualreviews.org • Forensic Seismology and the CTBT
229
ANRV374-EA37-10
ARI
27 March 2009
14:57
a
b
ABKAR
5
Vertical component 0.03–0.05 Hz
MDJ
WRA
0
–5
ASAR
Displacement (nm)
Annu. Rev. Earth Planet. Sci. 2009.37:209-236. Downloaded from arjournals.annualreviews.org
by University of British Columbia Library on 12/29/09. For personal use only.
nm at 1 Hz
ILAR
AKASG
GERES
40
INCN
0
–40
KS31
NVAR
0
PDAR
0
5
10
Time (s)
15
100
200
300
Time (s)
20
Figure 8
(a) Observed teleseismic P waves (blue) recorded globally, and (b) regional distance Rayleigh waves from DPRK 061009 (blue). The red
traces on the right are the synthetic seismograms for a point explosion source with a scalar moment of 2.0 × 1014 Newton-meters.
mb :Ms ratio is as expected for an explosion with small magnitude. The IDC experimental screening
line (shown in Figure 2) was devised under the assumption that mb and Ms scale differently with
yield, whereas an explosion model (Mueller & Murphy 1971) suggests that they should scale in
the same way. The inconsistency results from a failure to understand the overall relationship of
Ms to yield. Clearly, the theoretical basis of the mb :Ms criterion is not understood.
SUMMARY POINTS
1. The IMS is an important part of the CTBT verification regime. On completion, the IMS
network will comprise 337 facilities, 170 of which will be seismic stations.
2. The application of the analysis of seismic data to support verification of the Treaty has
been called forensic seismology and its main application is the identification of underground explosions from the thousands of earthquakes of potential interest that occur
each year.
3. There are four main tasks for the forensic seismologist: (a) signal detection, (b) association
of signals, (c) source location, and (d ) source identification.
230
Bowers
·
Selby
ANRV374-EA37-10
ARI
27 March 2009
14:57
4. There are four main complementary methods for source identification: (a) source depth,
(b) ratio of mb to Ms , (c) ratio of high-frequency (>2 Hz) P to S energy, and (d ) modelbased methods.
Annu. Rev. Earth Planet. Sci. 2009.37:209-236. Downloaded from arjournals.annualreviews.org
by University of British Columbia Library on 12/29/09. For personal use only.
5. The mb :Ms and high-frequency P/S criteria are empirically based on observations of
signals from presumed UNEs and earthquakes. Whereas models of the earthquake source
explain the observations reasonably well, current equivalent-elastic models of the UNE
do not.
6. Source-depth estimation is often difficult and unreliable, and model-based methods are
also subject to large uncertainties in the appropriate equivalent-elastic model for UNEs
fired in the diverse geological environments around the world.
7. Experience gained by forensic seismologists in the past decade at identifying suspicious
seismic sources using data from IMS and non-IMS stations suggests that although no
single method works all of the time, intelligent and original application of the complementary methods available is usually sufficient to satisfactorily identify the seismic source
in question.
8. Identifying a single underground explosion as nuclear or chemical is not possible using
seismology; diagnostic radionuclides must be detected. The IMS has a network of 80
radionuclide stations for this purpose, and there is provision for on-site inspection to
confirm violations of the Treaty.
FUTURE ISSUES
1. Development of equivalent-elastic models for UNEs in diverse geological media that
explain how explosions generate high-frequency P and S waves and long-period Rayleigh
and Love waves is necessary.
2. Development of realistic 3D models of the Earth (recent developments on resolving upper-mantle and crustal structure are described in this volume by M. Ritzwoller
in the review, “Earth Structure from Seismic Noise”) and the generation of synthetic seismograms from realistic earthquake and UNE sources in 3D models are
necessary.
3. It is important to identify the seismic source by solving the inverse problem using multiple
datasets (e.g., high-frequency P and S and long-period Rayleigh and Love waves), which,
if inverted, individually currently give contradictory solutions, presumably because the
models and/or parameters used are inappropriate.
DISCLOSURE STATEMENT
AWE Blacknest operates the National Data Center, on behalf of the United Kingdom National
Authority to the CTBT Organization, under contract with the Ministry of Defence (MoD). The
views expressed in this article are those of the authors and are not necessarily those of AWE, MoD,
or the United Kingdom government.
www.annualreviews.org • Forensic Seismology and the CTBT
231
ANRV374-EA37-10
ARI
27 March 2009
14:57
LITERATURE CITED
Annu. Rev. Earth Planet. Sci. 2009.37:209-236. Downloaded from arjournals.annualreviews.org
by University of British Columbia Library on 12/29/09. For personal use only.
Albright D, Gay C, Pabian F. 1999. New details emerge on Pakistan’s first nuclear test site. Earth Obs. Mag.
Jan. 7:No. 12
Anderson DN, Fagan DK, Tinker MA, Kraft GD, Hutchenson KD. 2007. A mathematical statistics formulation of the teleseismic explosion identification problem with multiple discriminants. Bull. Seism. Soc. Am.
97:1730–41
Anderson DN, Taylor SR. 2002. Application of regularized discrimination analysis to regional seismic event
identification. Bull. Seism. Soc. Am. 92:1078–90
Barker B, Clark M, Davis P, Fisk M, Hedlin M, et al. 1998. Monitoring nuclear tests. Science 281:1967–68
Barrientos S, Haslinger F, Astiz L, Bereza S, Given H, et al. 2001. Seismological monitoring of the Comprehensive Nuclear-Test-Ban Treaty. Kerntechnik 66:82–89
BBC. 2004a. Rumours linger over N Korea blast. http://news.bbc.co.uk/1/hi/world/asia-pacific/3652157.stm
(Accessed October 25, 2008)
BBC. 2004b. UK demands N Korea explain blast. http://news.bbc.co.uk/1/hi/uk politics/3648988.stm
(Accessed October 25, 2008)
Bennett TJ, McLaughlin KL. 1997. Seismic characteristics and mechanisms of rockbursts for use in seismic
discrimination. See Gibowicz & Lasocki 1997, pp. 61–66
Blandford RR. 1977. Discrimination between earthquakes and underground explosions. Annu. Rev. Earth
Planet. Sci. 5:111–22
Blandford RR. 1981. Seismic discrimination problems at regional distances. See Husebye & Mykkeltveit 1981,
pp. 695–740
Blandford RR. 1996. Regional seismic event discrimination. See Husebye & Dainty 1996, pp. 689–719
Bonner JL, Russell DR, Harkrider DG, Reiter DT, Herrmann RB. 2006. Development of a time-domain,
variable-period surface-wave magnitude measurement procedure for application at regional and teleseismic distances, Part II: application and Ms – mb performance. Bull. Seism. Soc. Am. 96:678–96
Bottone S, Fisk MD, McCartor GD. 2002. Regional seismic-event characterization using a Bayesian formulation of simple kriging. Bull. Seism. Soc. Am. 92:2277–96
Bowers D. 1997. The October 30, 1994 seismic disturbance in South Africa: earthquake or large rock burst?
J. Geophys. Res. 102:9843–57
Bowers D. 2002. Was the 16 August 1997 seismic disturbance near Novaya Zemlya an earthquake? Bull. Seism.
Soc. Am. 92:2400–9
Bowers D, Douglas A. 1997. Characterisation of large mine tremors using P observed at teleseismic distances.
See Gibowicz & Lasocki 1997, pp. 55–60
Bowers D, Douglas A. 1998. The effect of the earthquake radiation pattern on mb —a study using aftershocks
in the 1976 Gazli sequence. Bull. Seism. Soc. Am. 88:523–30
Bowers D, Douglas A, Selby ND, Marshall PD, Porter D, et al. 2002. Seismological identification of the 1998
May 28 Pakistan nuclear test. Geophys. J. Int.150:153–61
Bowers D, Marshall PD, Douglas A. 2001. The level of deterrence provided by data from the SPITS seismometer array to possible violations of the Comprehensive Test Ban in the Novaya Zemlya region. Geophys. J.
Int.146:425–38
Bowers D, Pearce RG, Douglas A. 2000. Using regional geophysical databases to facilitate the identification
of surface reflections on P-seismograms. Geophys. J. Int. 142:915–24
Bowers D, Walter WR. 2002. Discriminating between large mine collapses and explosions using teleseismic
P waves. Pure Appl. Geophys. 159:803–30
Bratt SR. 2001. The International Data Centre of the Comprehensive Nuclear-Test-Ban Treaty: vision and
progress. Kerntechnik 66:134–42
Chosun. 2004a. Ryongchon explosion eight times as great as North claims. http://english.chosun.com/
w21data/html/news/200405/200405160017.html (Accessed October 25, 2008)
Chosun. 2004b. Signs indicate no explosion occurred in N. Korea’s Kim Hyong-jik County. http://english.chosun.
com/w21data/html/news/200409/200409170025.html (Accessed October 25, 2008)
Christie DR, Vivas Veloso JA, Campus P, Bell M, Hoffmann T, et al. 2001. Detection of atmospheric nuclear
explosions: the infrasound component of the International Monitoring System. Kerntechnik 66:96–101
232
Bowers
·
Selby
Annu. Rev. Earth Planet. Sci. 2009.37:209-236. Downloaded from arjournals.annualreviews.org
by University of British Columbia Library on 12/29/09. For personal use only.
ANRV374-EA37-10
ARI
27 March 2009
14:57
CTBTO Prep. Comm. 2007. North Korea: a real test for the CTBT verification system? CTBTO Spectr. 9:24,
28
Denny MD, Johnson LR. 1991. The explosion seismic source function: Models and scaling laws reviewed.
See Taylor et al. 1991, pp. 1–24
Douglas A. 1981. Seismic source identification: a review of past and present research efforts. See Husebye &
Mykkeltveit 1981, pp. 1–48
Douglas A. 1997. Bandpass filtering to reduce noise on seismograms: Is there a better way? Bull. Seism. Soc.
Am. 87:770–77
Douglas A. 2007. Forensic seismology revisited. Surv. Geophys. 28:1–31
Douglas A, Bowers D, Marshall PD, Young JB, Porter D, et al. 1999. Putting nuclear-test monitoring to the
test. Nature 398:474–75
Douglas A, Bowers D, Young JB. 1997. On the onset of P seismograms. Geophys. J. Int. 129:681–90
Douglas A, Hudson JA, Blamey C. 1972. A quantitative evaluation of seismic signals at teleseismic distances—
III. Computed P and Rayleigh wave seismograms. Geophys. J. R. Astron. Soc. 28:385–410
Douglas A, Hudson JA, Kembhavi VK. 1971a. The analysis of surface wave spectra using a reciprocity theorem
for surface waves. Geophys J. R. Astron. Soc. 23:207–23
Douglas A, Hudson JA, Kembhavi VK. 1971b. The relative excitation of seismic surface and body waves by
point sources. Geophys. J. R. Astron. Soc. 23:451–60
Douglas A, Selby ND, Bowers D. 2002. Seismic magnitude and yield for the Indian nuclear test of 11 May
1998. Curr. Sci. 83:989–92
Ekström G, Richards PG. 1994. Empirical measurements of tectonic moment release in nuclear explosions
from teleseismic surface waves and body waves. Geophys. J. Int. 117:120–40
Evernden JF. 1998. Estimation of yields of underground explosions with emphasis on recent Indian and
Pakistani explosions. Phys. Soc. 27:10–11
Fan G-W, Lay T, Bottone S. 2002. Path corrections for source discriminants: a case study at two International
Monitoring System stations. Pure Appl. Geophys. 159:651–78
Fisk MD. 2006. Source spectral modeling of regional P/S discriminants at nuclear test sites in China and the
former Soviet Union. Bull. Seism. Soc. Am. 96:2348–67
Fisk MD. 2007. Corner frequency scaling of regional seismic phases for underground nuclear explosions at
the Nevada Test Site. Bull. Seism. Soc. Am. 97:977–88
Fisk MD, Gray HL, McCartor GD. 1996. Regional discrimination without transporting thresholds. Bull.
Seism. Soc. Am. 86:1545–58
Fisk MD, Jepsen D, Murphy JR. 2002. Experimental seismic event-screening criteria at the Prototype International Data Centre. Pure Appl. Geophys. 159:865–88
Fox BD, Woodhouse JH, Selby ND. 2005. Source depth determination in China using intermediate-period
fundamental mode surface waves. EOS Trans. AGU Fall Meet. Suppl. 86(52) (Abstr. S51B-1011)
Gibowicz SJ, Lasocki S, eds. 1997. Rockbursts and Seismicity in Mines. Rotterdam: Balkema. 437 pp.
Granville JP, Kim W-Y, Richards PG. 2002. An assessment of seismic body-wave magnitudes published by
the prototype International Data Centre. Seism. Res. Lett. 73:893–906
Granville JP, Richards PG, Kim W-Y, Sykes LR. 2005. Understanding the differences between three teleseismic mb scales. Bull. Seism. Soc. Am. 95:1809–24
Hartse HE. 1998. The 16 August 1997 Novaya Zemlya seismic event as viewed from GSN stations KEV and
KBS. Seism. Res. Lett. 69:206–15
Herrin E, Goforth T. 1986. Phase analysis of Rayleigh waves from the Shagan River test site in the USSR.
Bull. Seism. Soc. Am. 76:1739–54
Heyburn R, Bowers D. 2008. Earthquake depth estimation using the F trace and associated probability.
Geophys. J. Int. 98:18–35
Holzer TL, Fletcher JB, Fuis GS, Ryberg T, Brocher TM, Dietel CM. 1996. Seismograms offer insight into
Oklahoma City bombing. EOS Trans. Am. Geophys. Union 77:393, 398–99
Husebye ES, Dainty AM, eds. 1996. Monitoring a Comprehensive Test Ban Treaty. London: Kluwer. 836 pp.
Husebye ES, Mykkeltveit S, eds. 1981. Identification of Seismic Sources—Earthquake or Underground Explosion.
Holland: Reidel. 876 pp.
www.annualreviews.org • Forensic Seismology and the CTBT
233
ARI
27 March 2009
14:57
Jacob AWB, Willmore PL. 1972. Teleseismic P waves from a 10 ton explosion. Nature 236:305–6
Jenkins RD, Sereno TJ. 2001. Calibration of regional S/P amplitude-ratio discriminants. Pure Appl. Geophys.
158:1279–300
Karhu P, Clawson R. 2001. Radionuclide laboratories supporting the network of radionuclide stations in
verification of the Comprehensive Nuclear-Test-Ban Treaty. Kerntechnik 66:126–28
KCNA. 2004. Report on explosion at Ryongchon railway station. http://www.kcna.co.jp/item/2004/200404/
news04/24.htm#1 (Accessed October 25, 2008)
KCNA. 2006. DPRK successfully conducts underground nuclear test. http://www.kcna.co.jp/item/2006/200610/
news10/10.htm#1 (Accessed October 25, 2008)
Khristoforov B. 1996. About the control of the underwater and above water nuclear explosions by hydroacoustic
methods. Tech. Rep. SPC-95-4049. Russ. Acad. Sci., Moscow
Kim W-Y, Aharonian V, Lerner-Lam L, Richards PG. 1997. Discrimination of earthquakes and explosions
in Southern Russia using regional high-frequency three-component data from the IRIS/JSP Caucasus
network. Bull. Seism. Soc. Am. 87:569–88
Kim W-Y, Richards PG. 2007. North Korean nuclear test: seismic discrimination at low yield. EOS Trans.
Am. Geophys. Union 88:158, 161
Kim W-Y, Sykes LR, Armitage JH, Xie JK, Jacob KH, et al. 2001. Seismic waves generated by aircraft impacts
and building collapses at World Trade Center, New York City. EOS Trans. Am. Geophys. Union 82:565,
570–71
Koper KD, Wallace TC, Hollnack D. 1999. Seismic analysis of the 7 August 1998 truck bomb blast at the
American embassy in Nairobi, Kenya. Seism. Res. Lett. 70:512–21
Koper KD, Wallace TC, Reinke RE, Leverette JA. 2002. Empirical scaling laws for truck bomb explosions
based on seismic and acoustic data. Bull. Seism. Soc. Am. 92:527–42
Koper KD, Wallace TC, Taylor SR, Hartse HE. 2001. Forensic seismology and the sinking of the Kursk.
EOS Trans. Am. Geophys. Union 82:37, 45–46
Kradolfer U. 2000. Waves4U: waveform availability through AutoDRM’s. Seism. Res. Lett. 71:79–82
Lawrence M, Galindo M, Grenard P, Newton J. 2001. Hydroacoustic monitoring system for the Comprehensive Nuclear-Test-Ban Treaty. Kerntechnik 66:90–95
Lilwall RC. 1987. Station threshold bias in short-period amplitude distance and station terms used to compute
body-wave magnitudes mb . Geophys. J. R. Astron. Soc. 91:1127–33
Lilwall RC. 1988. Regional mb :Ms , Lg /Pg amplitude ratios and Lg spectral ratios as criteria for distinguishing
between earthquakes and explosions: a theoretical study. Geophys. J. 93:137–47
Marshall PD, Basham PW. 1972. Discrimination between earthquakes and underground explosions employing
an improved Ms scale. Geophys. J. R. Astron. Soc. 28:431–58
Marshall PD, Rozgonova D, Weiss W. 2001. Testing times: the road to a Comprehensive Nuclear-Test-Ban
Treaty. Kerntechnik 66:80–81
Marshall PD, Springer DL, Rodean HC. 1979. Magnitude corrections for attenuation in the upper mantle.
Geophys. J. R. Astron. Soc. 57:609–38
Matthews M, Schulze J. 2001. The radionuclide monitoring system of the Comprehensive Nuclear-Test-Ban
Treaty Organisation: from sample to product. Kerntechnik 66:102–11
McCormack DA. 2003. Using seismic data in air crash investigations. CTBTO Spectr. 2:1, 16
Mueller RA, Murphy JR. 1971. Seismic characterization of underground nuclear detonations. Part I. Seismic
scaling law of underground nuclear detonations. Bull. Seism. Soc. Am. 61:1675–92
Murphy JR. 1996. Types of seismic event and their source descriptions. See Husebye & Dainty 1996,
pp. 225–45
Murphy JR, Barker BW. 2006. Improved focal-depth determination through automated identification of the
seismic depth phases pP and sP. Bull. Seism. Soc. Am. 96:1213–29
Myers SC, Wagoner J, Larsen S, Rodgers A, Mayeda K, et al. 2003. Simulation of regional explosion Sphases (SiRES) project. Proc. 2003 Seism. Res. Rev., Tucson, AZ, Tech. Rep. UCRL-JC-153604. pp. 117–24.
Livermore, CA: Lawrence Livermore Natl. Lab.
Myers SC, Walter WR, Mayeda K, Glenn L. 1999. Observations in support of Rg scattering as a source for
explosion S waves: regional and local recordings of the 1997 Kazakhstan depth of burial experiment.
Bull. Seism. Soc. Am. 89:544–49
Annu. Rev. Earth Planet. Sci. 2009.37:209-236. Downloaded from arjournals.annualreviews.org
by University of British Columbia Library on 12/29/09. For personal use only.
ANRV374-EA37-10
234
Bowers
·
Selby
Annu. Rev. Earth Planet. Sci. 2009.37:209-236. Downloaded from arjournals.annualreviews.org
by University of British Columbia Library on 12/29/09. For personal use only.
ANRV374-EA37-10
ARI
27 March 2009
14:57
Natl. Acad. Sci. 2002. Technical issues related to the Comprehensive Nuclear Test Ban Treaty. Washington,
DC: Natl. Acad. Press. 84 pp.
Patton HJ. 1991. Seismic moment estimation and the scaling of the long-period explosion source spectrum.
See Taylor et al. 1991, pp. 171–83
Patton HJ. 1998. Bias in the centroid moment tensor for central Asian earthquakes: evidence from regional
surface wave data. J. Geophys. Res. 103:26963–74
Pearce RG. 1980. Fault plane solutions using relative amplitudes of P and surface reflections: further studies.
Geophys. J. R. Astron. Soc. 60:459–87
Rezapour M, Pearce RG. 1998. Bias in surface wave magnitude Ms due to inadequate distance corrections.
Bull. Seism. Soc. Am. 88:43–61
Richards PG, Kim W-Y. 1997. Testing the nuclear test-ban treaty. Nature 389:781–82
Richards PG, Zavales J. 1996. Seismological methods for monitoring a CTBT: the technical issues arising in
early negotiations. See Husebye & Dainty 1996, pp. 53–81
Ringdal F, Kremenetskaya E, Asming V. 2001. Observed characteristics of regional seismic phases and implications for P/S discrimination in the European Arctic. Pure Appl. Geophys. 159:701–19
Ringdal F, Kværna T, Paulsen B. 2000. Seismic events in the Barents Sea at and near the site of the Kursk
submarine accident on 12 August 2000. NORSAR Sci. Rep. 1-2000/1002, Norway, pp. 77–88
Rodgers AJ, Walter WR. 2002. Seismic discrimination of the May 11, 1998 Indian nuclear test with shortperiod regional data from station NIL (Nilore, Pakistan). Pure Appl. Geophys. 159:679–700
Rodgers AJ, Walter WR, Bredbeck TM. 1999a. Complete regional waveform modelling to estimate seismic
velocity structure and source parameters for CTBT monitoring. Proc. 21st Seism. Res. Symp., Las Vegas,
NV, Tech. Rep. LA-UR-99-4700. pp. 226–33. Los Alamos, NM: Los Alamos Natl. Lab.
Rodgers AJ, Walter WR, Schultz CA, Myers SC, Lay T. 1999b. A comparison of methodologies for representing path effects on regional P/S discriminants. Bull. Seism. Soc. Am. 89:394–408
Russell D. 2006. Development of a time-domain, variable-period surface wave magnitude measurement procedure for application at regional and teleseismic distances, part I: theory. Bull. Seism. Soc. Am. 96:665–77
Rygg E. 1979. Anomalous surface waves from underground explosions. Bull. Seism. Soc. Am. 69:1995–2002
Schultz CA, Myers SC, Hipp J, Young CJ. 1998. Nonstationary Bayesian kriging: a predictive technique
to generate spatial corrections for seismic detection, location, and identification. Bull. Seism. Soc. Am.
88:1275–88
Schweitzer J. 2002. Some results derived from the seismic signals of the accident of the Russian submarine
Kursk. NORSAR Sci. Rep. 1-2002, Norway, pp. 115–21
Sebe O, Bard P-Y, Guilbert J. 2005. Single station estimation of seismic source time function from coda waves:
The Kursk disaster. Geophys. Res. Lett. 32:L14308
Selby ND. 2001. Association of Rayleigh waves using back-azimuth measurements: application to test ban
verification. Bull. Seism. Soc. Am. 91:580–93
Selby ND, Bowers D. 2007. Implications of the 9 October 2006 North Korean nuclear test for event screening.
Seism. Res. Lett. 78:253
Selby ND, Bowers D, Douglas A, Heyburn R, Porter D. 2005. Seismic discrimination in Southern Xinjiang:
The 13 March 2003 Lop Nor earthquake. Bull. Seism. Soc. Am. 95:197–211
Selby ND, Bowers D, Marshall PD, Douglas A. 2003. Empirical path and station corrections for surface-wave
magnitude, Ms , using a global network. Geophys. J. Int. 155:379–90
Selby ND, Walter WR, Bowers D. 2004. Tectonic release triggered by the 11 May 1998 Indian nuclear tests.
Seism. Res. Lett. 74:252
Sikka SK, Roy F, Nair GJ. 2001. Yield estimation of Indian nuclear tests of 1998. Curr. Sci. 81:885–87
Stevens JL, Baker GE, Xu H. 2006. The physical basis of the explosion source and generation of regional
seismic phases. Proc. 28th Seism. Res. Rev., Orlando, Fla., Tech. Rep. LA-UR-06-5471. pp. 681–92. Los
Alamos, NM: Los Alamos Natl. Lab.
Stevens JL, Day SM. 1985. The physical basis of mb :Ms and variable frequency magnitude methods for
earthquake/explosion discrimination. J. Geophys. Res. 90:3009–20
Stevens JL, McLaughlin KL. 2001. Optimization of surface wave identification and measurement. Pure Appl.
Geophys. 158:1547–82
www.annualreviews.org • Forensic Seismology and the CTBT
235
ARI
27 March 2009
14:57
Stevens JL, Murphy JR. 2002. Yield estimation from surface-wave amplitudes. Pure Appl. Geophys. 158:2227–51
Sykes L. 2002. Four decades of progress in seismic identification help verify the CTBT. EOS Trans. Am.
Geophys. Union 83:497, 500
Takano M, Krioutchenkov V. 2001. Technical methods employed for the On-Site Inspection. Kerntechnik
66:143–46
Taylor SR, Patton HJ, Richards PG, eds. 1991. Explosion Source Phenomenology. Geophys. Monogr. 65.
Washington, DC: Ser. Am. Geophys. Union. 268 pp.
Taylor SR, Velasco AA, Hartse HE, Phillips WS, Walter WR, et al. 2002. Amplitude corrections for regional
seismic discriminants. Pure Appl. Geophys. 159:623–50
Toksöz MN, Kehrer HH. 1972. Tectonic release by underground nuclear explosions and its effect on seismic
discrimination. Geophys. J. R. Astron. Soc. 31:141–61
Tsai Y-B, Aki K. 1970. Precise focal depth determination from amplitude spectra of surface waves. J. Geophys.
Res. 75:5729–43
van der Vink GE, Park J. 1994. Nuclear test ban monitoring: new requirements, new resources. Science
263:634–35
Walter WR, Matzel E, Pasyanos ME, Harris DB, Gok R, et al. 2007. Empirical observations of earthquakeexplosion discrimination using P/S ratios and implications for the sources of explosion S-waves. Proc. 29th
Monit. Res. Rev., Denver, Colo., Tech. Rep. LA-UR-07-5613. pp. 684–93. Los Alamos, NM: Los Alamos Natl.
Lab.
Walter WR, Mayeda KM, Patton HJ. 1995. Phase and spectral ratio discrimination between NTS earthquakes
and explosions. Part I: empirical observations. Bull. Seism. Soc. Am. 85:1050–67
Walter WR, Taylor SR. 2001. A revised magnitude and distance amplitude correction (MDAC2) procedure
for regional seismic discriminants: Theory and testing at NTS. Tech. Rep. UCRL-ID-146882. Lawrence
Livermore Natl. Lab., Livermore, CA
Xie J, Patton HJ. 1999. Regional phase excitation and propagation in the Lop Nor region of central Asia and
implications for P/Lg discriminants. J. Geophys. Res. 104:941–54
Zaslavsky-Paltiel I, Steinberg DM. 2008. Comparison of methods for estimating station magnitude corrections
for improved seismologic monitoring of the Comprehensive Nuclear-Test-Ban Treaty. Bull. Seism. Soc.
Am. 97:1–17
Zhang J, Lay T, Zaslow J, Walter WR. 2002. Source effects on regional seismic discriminant measurements.
Bull. Seism. Soc. Am. 91:2926–45
Annu. Rev. Earth Planet. Sci. 2009.37:209-236. Downloaded from arjournals.annualreviews.org
by University of British Columbia Library on 12/29/09. For personal use only.
ANRV374-EA37-10
RELATED RESOURCES
http://www.ctbto.org. CTBT Organisation Web site
http://www.iss.niiit.ru/ksenia/catal nt/index.htm. Catalog of worldwide nuclear testing
236
Bowers
·
Selby
AR374-FM
ARI
27 March 2009
18:4
Annual Review
of Earth and
Planetary Sciences
Annu. Rev. Earth Planet. Sci. 2009.37:209-236. Downloaded from arjournals.annualreviews.org
by University of British Columbia Library on 12/29/09. For personal use only.
Contents
Volume 37, 2009
Where Are You From? Why Are You Here? An African Perspective
on Global Warming
S. George Philander ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! 1
Stagnant Slab: A Review
Yoshio Fukao, Masayuki Obayashi, Tomoeki Nakakuki,
and the Deep Slab Project Group ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !19
Radiocarbon and Soil Carbon Dynamics
Susan Trumbore ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !47
Evolution of the Genus Homo
Ian Tattersall and Jeffrey H. Schwartz ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !67
Feedbacks, Timescales, and Seeing Red
Gerard Roe ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !93
Atmospheric Lifetime of Fossil Fuel Carbon Dioxide
David Archer, Michael Eby, Victor Brovkin, Andy Ridgwell, Long Cao,
Uwe Mikolajewicz, Ken Caldeira, Katsumi Matsumoto, Guy Munhoven,
Alvaro Montenegro, and Kathy Tokos ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! 117
Evolution of Life Cycles in Early Amphibians
Rainer R. Schoch ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! 135
The Fin to Limb Transition: New Data, Interpretations, and
Hypotheses from Paleontology and Developmental Biology
Jennifer A. Clack ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! 163
Mammalian Response to Cenozoic Climatic Change
Jessica L. Blois and Elizabeth A. Hadly ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! 181
Forensic Seismology and the Comprehensive Nuclear-Test-Ban Treaty
David Bowers and Neil D. Selby ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! 209
How the Continents Deform: The Evidence from Tectonic Geodesy
Wayne Thatcher ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! 237
The Tropics in Paleoclimate
John C.H. Chiang ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! 263
vii
AR374-FM
ARI
27 March 2009
18:4
Rivers, Lakes, Dunes, and Rain: Crustal Processes in Titan’s
Methane Cycle
Jonathan I. Lunine and Ralph D. Lorenz ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! 299
Planetary Migration: What Does it Mean for Planet Formation?
John E. Chambers ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! 321
The Tectonic Framework of the Sumatran Subduction Zone
Robert McCaffrey ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! 345
Annu. Rev. Earth Planet. Sci. 2009.37:209-236. Downloaded from arjournals.annualreviews.org
by University of British Columbia Library on 12/29/09. For personal use only.
Microbial Transformations of Minerals and Metals: Recent Advances
in Geomicrobiology Derived from Synchrotron-Based X-Ray
Spectroscopy and X-Ray Microscopy
Alexis Templeton and Emily Knowles ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! 367
The Channeled Scabland: A Retrospective
Victor R. Baker ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! 393
Growth and Evolution of Asteroids
Erik Asphaug ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! 413
Thermodynamics and Mass Transport in Multicomponent, Multiphase
H2 O Systems of Planetary Interest
Xinli Lu and Susan W. Kieffer ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! 449
The Hadean Crust: Evidence from >4 Ga Zircons
T. Mark Harrison ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! 479
Tracking Euxinia in the Ancient Ocean: A Multiproxy Perspective
and Proterozoic Case Study
Timothy W. Lyons, Ariel D. Anbar, Silke Severmann, Clint Scott,
and Benjamin C. Gill ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! 507
The Polar Deposits of Mars
Shane Byrne ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! 535
Shearing Melt Out of the Earth: An Experimentalist’s Perspective on
the Influence of Deformation on Melt Extraction
David L. Kohlstedt and Benjamin K. Holtzman ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! 561
Indexes
Cumulative Index of Contributing Authors, Volumes 27–37 ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! 595
Cumulative Index of Chapter Titles, Volumes 27–37 ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! 599
Errata
An online log of corrections to Annual Review of Earth and Planetary Sciences articles
may be found at http://earth.annualreviews.org
viii
Contents